Search papers, labs, and topics across Lattice.
1
0
3
Results show that scaling consistently improves performance, while multitask learning acts as a beneficial regularizer primarily for smaller-capacity models.
